Transcript of "AGGLOMERATION AND INTERREGIONAL MOBILITY OF LABOR IN PORTUGAL"
AGGLOMERATION AND INTERREGIONAL MOBILITY OF LABOR IN PORTUGAL Vitor João Pereira Domingues Martinho Unidade de I&D do Instituto Politécnico de Viseu Av. Cor. José Maria Vale de Andrade Campus Politécnico 3504 - 510 Viseu (PORTUGAL) e-mail: firstname.lastname@example.org ABSTRACT The aim of this paper is to analyze the relationship between inter-industry, intra-industry and inter-regional clustering and demand for labor by companies in Portugal. Isexpected at the outset that there is more demand for work where the agglomeration is greater.It should be noted, as a summary conclusion, the results are consistent with the theoreticaldevelopments of the New Economic Geography, namely the demand for labor is greaterwhere firms are better able to cluster that is where transport costs are lower and where there isa strong links "backward and forward" and strong economies of agglomeration. Keywords: agglomeration; Portuguese regions; labor; interregional mobility. 1. INTRODUCTION Related to the agglomeration process much has been writing and in varied points ofview, that is, in some cases on their causes and others about the process itself. Works likethose of (1)Florence (1948), (2)Perloff et al. (1960), (3)Fuchs (1962), (4)Krugman (1991) and(5)Dumais et al. (1997), for example, have sought to investigate the phenomenon ofgeographical concentration of economic activity, based on the costs of transport andcommunication.
In this work, however, we will focus on the relationship between agglomeration andregional demand for labor in industry, analyzing the effect of three forces of agglomeration,transport costs that encourage companies to put up with their activities in regions with relativelow cost access to foreign markets; linkage "backward and forward" that give companies anincentive to put near their buyers and suppliers; and economies of agglomeration effects("spillover") that tend to reinforce the concentration of economic activity, ie, companiesbenefit from being together, even if they have business relationships with each other and canbenefit from, for example, experience accumulated by other companies. It is consideredtherefore that these three forces create the conditions for which there is agglomeration,through processes of growth pattern circular and cumulative, and that where there isovercrowding there is greater demand for labor by firms (6)(Hanson, 1998). For transport costs, Krugman (1991), for example, showed that the interaction betweeneconomies of scale at the firm level and transportation costs can explain the formation ofcities. (7)Krugman and Livas (1992), on the other hand, showed that the size and location ofcities is conditioned by the degree of openness of the entire economy. Finally, (8)Rauch(1993) found that transportation costs determine the volume of trade within the region andbetween countries. Transport costs help to inquire about the inter-regional agglomeration,since one of the purposes of the agglomeration of population and economic activity in a givenregion is the cost savings of transport and communication. For the linkages "backward and forward" considered in this work, to stress the fact that(9)Venables (1996) and (10)Krugman and Venables (1995), have formalized the concepts of(11)Hirschman (1958), where the vertical relationships between industries create a model ofinterdependence the location of industry. Following this expansion of an industry contributesto the expansion of other industries in the same place, which are connected with thiscommercially, either upstream or downstream, i.e., companies who trade with each otherintermediate goods benefit from the proximity factor, a time, save on transportation costs andcommunication. So this is a more simplistic way of looking this up, i.e. only betweenindustries, because in much of the works related to economic geography the links "backwardand forward", representing the centripetal forces, are considered in a more global way and notonly between commercially related industries upstream and downstream. With theseconnections "backward and forward" is intended to analyze the inter-industry clustering. To reinforce what has been stated previously, noted also that recent theories about thelocation of economic activity have three elements in common, increasing returns to scale,transportation costs and communication costs and congestion. The new aspect of the industry
cluster in this work relates to the fact that companies share links "backward and forward" toeach other in the exchange of intermediate goods, as previously mentioned, the centripetalforces, and not only with employees who are also consumers, as usually stated in most of thework on these issues. 2. THE ANALYTICAL MODEL OF THE RELATIONSHIP BETWEENAGGLOMERATION AND REGIONAL DEMAND FOR LABOR To analyze the relationship between inter-industry clustering, intra-industry and inter-regional and regional demand for labor, we tried to develop an equation that relates thedemand relative for labor regional by the companies (represented by employment in eachmanufacturing Portuguese in a given region of the country, of overall employment in thisindustry in Portugal) with the three explanatory factors mentioned above the agglomeration,or transportation costs and communication, links "backward and forward" and agglomerationeconomies, following the procedures of Hanson (1998). Thus, we obtained the following equation, where we assume that there are positivetransportation costs which take the form "iceberg" of Samuelson, i.e. each unit of output ofregion i sent to another, only a fraction reaches the T destination. The Lij is employment inregion i in industry j, and Rem is the salary. L Re mijt 1 / Lijt 1 Tij L /L L /L ln ijt 0 1 ln 2 ln 3 ln ikt 1 ijt 1 4 ln ijt 1 it 1 L Re m / L ijtTij L /L L /L jt jt 1 jt 1 i kt 1 jt 1 jt 1 t 1 , Liht 1 / Lit 1 2 5 ln h j 2 ijt jt Lht 1 / Lt 1 h j equation of the relative growth of employment. Specified in this equation is the relative growth of employment, i.e. employmentgrowth at a Portuguese manufacturing industry in a region of overall employment in thisindustry in Portugal, which represents the relative regional labor demand by the Portugueseindustrial firms. This relative demand for labor is a function of initial conditions in eachregional industry in relation to the national industry, where i is the index of the region, j of
each manufacturing, operating in Portugal, considered in the sample used, k the totalmanufacturing and h the total manufacturing except j. Considered to be the explanatoryvariables lagged one period to analyze the influence of each factor present in a given period insearch of work the following year and to avoid introducing bias due to simultaneity. 3. STATISTICAL DATA USED Taking into account the variables of the model presented previously, we used temporalstatistical data of the five regions of mainland Portugal, from the regional database of Eurostatstatistics (Eurostat Regio of Statistics 2000), relating to salaried employees, regional andnational in the manufacturing sector and overall economic activity, the nominal wages inmanufacturing and the flow of goods from each of the regions of mainland Portugal to Lisboae Vale do Tejo. It is this region from the outset as a potential place of agglomeration, given itscharacteristics, namely, the fact that the highest salaries, to be regions where there are moreemployees in manufacturing and the region to be associated with higher flows of goods.Nominal wages are solely those of the manufacturing industry, given the emphasis that isgiven to the sector of manufactured goods, since it is the sectors that produce tradable goodsmostly. The regional flow of goods intended to be a "proxy" to transport costs, given that thisis an indirect way to measure them, as the authors admit this theoretical approach. In the faceof it was considered disaggregated data for nine manufacturing industries operating inPortugal and a series from 1986 to 1994, we had a total of 405 observations in the panel. 4. THE ESTIMATES MADE In all estimations made possible with panel data, the best results, according to thetheory, are obtained in the estimations with variables "dummies", with differences andrandom effects estimates are given in the table below. Were considered 45 "dummies", onefor each individual, since the data relate to five regions and nine manufacturing industries. InTable 1 each line refers to an industry for the five regions considered, and each column refersto a region for the nine industries, always in the order mentioned above. Thus, the sixth line of"dummies" refers to food that is not used for worsen statistically the estimation results,possibly because it was an industry with specific features, since the amount depends onagriculture.
However, the Hausman test indicates that the best results are the estimates of fixedeffects. Anyway, we present the results obtained in the three estimations, even to serve as acomparison. Explore will be, above all, the results obtained in the estimations with variables"dummies" because they were statistically more satisfactory. It should be noted also that thedependent variable was not considered in growth rate, since this way the results werestatistically weaker. Were also carried out simulations considering the variables in theequation productivity and unemployment, but worsen the results statistically and induce theappearance of strange signs for the coefficients, we chose not to consider. Table 1: Estimation of the equation for employment LSDV(1) D(2) GLS(3) 0 -1.878(5)* (-4.424) (6) D1(4) D2(4) D3(4) D4(4) D5(4) (5) (5) (5) (5) -0.970 -1.575 -1.470 * 6.782 (b) (6) (6) (6) (6) (-0.379) (-0.616) (-0.575) (2.579) D6(4) D7(4) D8(4) D9(4) D10(4) -0.620(5) -0.576(5) -1.910(5) -8.344(5)* -4.204(5) (6) (6) (6) (6) (-0.242) (-0.225) (-0.748) (-3.229) (-1.634) (6) D11(4) D12(4) D13(4) D14(4) D15(4) (5) (5) (5) (5) 1.967 -3.757 -1.344 -0.100 -10.038(5)* (0.768) (6) (-1.465) (6) (-0.526) (6) (-0.039) (6) (-2.553) (6) (4) (4) (4) (4) D16 D17 D18 D19 D20(4) 0.658(5) 0.792(5) -3.394(5) -4.587(5)** -2.234(5) (6) (6) (6) (6) (0.257) (0.310) (-1.327) (-1.763) (-0.868) (6) D21(4) D22(4) D23(4) D24(4) D25(4) (5) (5) (5) (5) 0.986 -0.780 -2.052 0.149 -1.317(5) (0.385) (6) (-0.305) (6) (-0.803) (6) (0.058) (6) (-0.512) (6) (4) (4) (4) (4) D26 D27 D28 D29 D30(4) (c) (c) (c) (c) (c) D31(4) D32(4) D33(4) D34(4) D35(4) 0.223(5) -2.587(5) -3.329(5) -8.367(5)* -13.384(5)* (6) (6) (6) (6) (0.087) (-1.009) (-1.303) (-3.240) (-3.314) (6) D36(4) D37(4) D38(4) D39(4) D40(4) (5) (5) (5) (5) -0.073 -0.685 -1.420 -5.311 * -1.921(5) (-0.029) (6) (-0.268) (6) (-0.556) (6) (-2.049) (6) (-0.747) (6) (4) (4) (4) (4) D41 D42 D43 D44 D45(4) 0.881(5) -1.352(5) -3.327(5) -6.699(5)* -7.784(5)* (6) (6) (6) (6) (0.344) (-0.529) (-1.302) (-2.591) (-3.025) (6) 1 1 1 (5) (5) 0.119 * 0.122 * 0.112(5)* (6) (6) (2.086) (2.212) (1.973) (6)
2 2 2 0.018(5)** 0.012(5) 0.022(5)* (1.879) (6) (1.286) (6) (2.179) (6) 3 3 3 (5) (5) 1.301 * 1.127 * 0.979(5)* (6) (6) (3.241) (2.918) (2.443) (6) 4 4 4 (5) (5) 0.731 * 0.661 ** 0.549(5) (1.989) (6) (1.867) (6) (1.492) (6) 5 5 5 (5) (5) -0.759 * -0.744 * -0.581(5)* (6) (6) (-4.357) (-4.525) (-3.401) (6)R2 adjusted 0.987 0.217 0.683Durbin-Watson 2.298 2.086 2.068Hausman TestChi-square 7777.548*(a)(1) Estimation with 45 variables "dummies", one for each manufacturing industry, (2) Estimation withdifferences, (3) Estimation with random effects, (4) Variables "Dummies" (5) value of the coefficient, (6) T- statistic * coefficient statistically significant at the 5% level, ** coefficient statistically significant at 10%(a) reject the hypothesis of random effects, (b) not considered this "dummy" values to present strangers:(c) Do not consider these "dummies" by statistically worse results For variable coefficients "dummies" to mention that vary in terms of value andsignificance, particularly between regions, in particular, for the last two regions that are theAlentejo and the Algarve. This means that these two regions have significant differences ineconomic structures relatively to the other three, namely, that the Alentejo, because thisregion has little economic activity and the Algarve have, especially tourism. On the other hand, the equation for employment (employment in each Portuguesemanufacturing industry in a given region of the country for employment in this industry inPortugal) gives satisfactory results in terms of statistical significance of the coefficients of the"no dummies," the degree of accuracy of adjustment and autocorrelation of errors. Tohighlight the fact that almost all coefficients of the variables "no dummy" with an elasticityless than unity, with the exception of links to the "backward and forward", which indicates theimportance of these linkages in explaining the relative employment. Analyzing the results ofthe estimation for the variables "no dummy" there is, as might be expected, given thedevelopments of the New Economic Geography, there is a positive relationship betweenrelative employment and relative nominal wages (salaries regional industrial concerning thenational industrial wages), the positive effect is confirmed also in relation to the links"backward and forward" (ratio between the number of employees in total manufacturing in
each region and the number of employees in each manufacturing considered in this region forthe same ratio at national level) and in relation to economies of agglomeration (ratio betweenthe number of employees in each manufacturing industry in a given region and the totalnumber of employees in all economy of that region, on the same ratio found at national level).On the other hand, confirms the negative relationship between transport costs and demand onemployment (assuming that costs of transport and cargo flows vary inversely) and the ratio ofthe distribution of employment across industries (the ratio of the sum square of the number ofemployees in total manufacturing (other than that is being analyzed) of a given region and thetotal number of employees throughout the economy of this region, for the same sumconsidered at national level). Note, finally, that all coefficients are statistically significant to5% and 10% and the high value of the degree of accuracy of adjustment. 5. CONCLUSIONS Given the statistical analysis performed, it appears that for the regions, we can safelysay is that on the one hand, the Norte specializes practically in the textile industry and,second, Lisboa e Vale do Tejo has a wide variety of industries, but has the best values for theexplanatory factors of regional demand for labor in industry.Complementing the analysis of the data and the estimation results, noted also thattransportation costs are important because, in addition to the statistical significance of thecoefficient associated with this variable, the regions closer to the Lisboa e Vale do Tejo hasthe greater flow of goods to this region. There are also links "backward and forward" betweenthe different manufacturing industries, there are economies of agglomeration effects("spillover") between firms and industries and there is a distribution, more or less uniform, ofthe employment across different industries. Productivity and unemployment have noinfluence in explaining the demand for industrial work at the regional level, given the resultsobtained in the estimations, when these variables were considered. Following this, it appears that the results are consistent with the developments of theNew Economic Geography which emphasizes the transportation costs, which are based on aset of other explanatory variables, such as links "backward and forward," i.e., buyers andsuppliers are looking to be together in order to save on transport costs and agglomerationeconomies. Note that Alfred Marshall in 1920 had made reference to such connections, whenthe modeling of increasing returns that explain the spatial concentration, has proposed athreefold classification that is presented below. In modern terminology, he advocated that the
locations appear on the face of industrial purposes "spillovers," the advantages of specializedmarkets and links "backward" and "forward" associated with large local markets. Although allthese three forces are clearly operating in the real world, the New Economic Geography hasgenerally ignored the first two, mainly because they are difficult to model in an explicitmanner. Finally, in light of the foregoing, it is noted that Lisboa e Vale do Tejo is a localpotential agglomeration of population and economic activity (as we suppose), since the flowof goods is higher in regions closer to this region , relative employment is higher in theprevious period where wages were higher (knowing that it is in Lisboa e Vale do Tejo thatwages are higher) and to characterize inter-industry and inter-industry described. There are other factors which, although implicit, would be worthwhile to disaggregatefurther analysis, as the entrepreneurial dynamics ((12)Martinho, 2010a and (13)Martinho,2010b) and the spatial effects ((14)Martinho, 2011a, and (15)Martinho, 2011b). 6. REFERENCES1. S. Florence. Investment, location and size of plant. Cambridge University Press, 1948.2. H. Perloff; E. Dunn; E. Lampard and R. Muth. Regions, resources and economic growth.Johns Hopkins Press, 1960.3. V. Fuchs. Changes in the location of manufacturing in the United States since 1929. YaleUniversity Press, 1962.4. P. Krugman. Increasing returns and economic geography. Journal of Political Economy(99), 483-499 (1991).5. G. Dumais; G. Ellison and E. Glaeser. Geographic concentration as a dynamic process.Mimeo, 1997.6. G. Hanson. Regional adjustment to trade liberalization. Regional Science and UrbanEconomics (28), 419-444 (1998).7. P. Krugman and, R. Livas. Trade policy and the third world metropolis. NBER WorkingPaper Nº 4238, 1992.8. J.E. Rauch. Comparative advantage, geographic advantage and the volume of trade.Economic Journal (101), 1230-1244 (1993).9. A.J. Venables. Equilibrium locations of vertically – linked industries. InternationalEconomic Review (37), 341-360 (1996).
10. P. Krugman and A.J. Venables. Globalization and the inequality of nations. QuarterlyJournal of Economics (110), 857-880 (1995).11. A.O. Hirschman. The strategy of economic development. Yale University Press, NewHaven, 1958.12. V.J.P.D. Martinho. Entrepreneurship: what´s happening?. Revista Millenium, nº39, IPV,(2010a);13. V.J.P.D. Martinho. Entrepreneurship: some considerations. Revista Millenium, nº39, IPV,(2010b);14. V.J.P.D. Martinho. Spatial Effects and Convergence Theory in the Portuguese Situation.International Journal and Academic Research, Volume 3. Issue 3. May 30 (2011a);15. V.J.P.D. Martinho. Spatial Effects and Verdoorn Law in the Portuguese Context.International Journal and Academic Research, Volume 3. Issue 3. May 30 (2011b).